Regional landslide parameter constraint and uncertainty quantification on the Qinghai-Tibet Plateau using a geographically-enabled bayesian framework and paleo-landslide data

The Qinghai-Tibet Plateau (QTP) is highly susceptible to large-scale landslides. Reliable regional hazard assessment is challenged by data scarcity and parameter uncertainty. This study addresses these issues by developing a ‘Geographically-Enabled Bayesian Framework’ and applying it to a unique dataset of 42 paleo-landslides on the QTP. The framework synergistically integrates geomatics (SRTM DEM-based topography reconstruction), physics-based simulation (Massflow), and advanced statistical modeling. Using Gaussian Process Regression surrogates and an adaptive MCMC algorithm, the framework efficiently constrains two key mobility parameters—the basal friction coefficient (μ) and pore water pressure ratio (ru). The framework’s performance is demonstrated by a marked reduction in parameter uncertainty (average standard deviation decrease: 24.5% for μ and 36.7% for ru), providing robust, data-driven constraints. Its robustness is further substantiated through a preliminary cross-regional validation, confirming generalizability. This study also discusses how geospatial data uncertainty influences results and how constrained parameters can inform risk mitigation. Ultimately, this work delivers a validated, scalable methodology and crucial parameter estimates, establishing a more rigorous foundation for landslide hazard and risk assessment on the QTP.